2053: Current State and Completeness of Reporting Clinical Prediction Models Using Machine Learning in Systemic Lupus Erythematosus: A Systematic Review

Claudia Mendoza-Pinto¹, Ivet Etchegaray-Morales², Pamela Munguía-Realpozo², David Angel Osorio-Peña², Socorro Méndez-Martínez¹ and Mario Garcia-Carrasco², ¹Instituto Mexicano del Seguro Social, Puebla, Mexico, ²Benemérita Universidad Autónoma de Puebla, Puebla, Mexico

Background/Purpose: There is increased interest in machine learning (ML)-based prediction models in systemic lupus erythematosus (SLE). We made a systematic review of adherence in diagnostic and prognostic applications of ML in SLE using the Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD) Statement.

Methods: A systematic review using 5 databases was conducted from inception until December 2021. We identified studies that examined the application of ML for prognosis and/or diagnostic purposes. We used the Preferred Reporting Items for Systematic Review and Meta-analysis (PRISMA) in the search and screening processes. Adherence to reporting standards was assessed by TRIPOD.

Results: Our search identified 385 studies, of which 44 were included, 28 (63.6%) diagnostic and 16 (36.4%) prognostic prediction model studies. Overall, article adhered between 17% and 67% (median 43%, IQR 37-49%) to TRIPOD items. No article fully adhered to complete reporting of the abstract, and very few reported the model's predictive performance (2.3%, 95% CI 0.06 to 12.0), testing of interaction terms (2.3%, 95% CI 0.06 to 12.0), flow of participants (50%, 95% CI, 34.6 to 65.4), blinding of predictors (2.3%, 95% CI 0.06 to 12.0), handling of missing data (36.4%, 95% CI 22.4 to 52.2) and appropriate title (20.5%, 95% CI 9.8 to 35.3). There was often almost complete reporting of the source of data (88.6%, 95% CI 75.4 to 96.2), eligibility criteria (86.4%, 95% CI 76.2 to 96.5) and interpretation of results (88.6%, 95% CI 75.4 to 96.2).

Conclusion: The completeness of reporting of diagnostic and prognostic prediction model studies using ML in SLE was poor. Several items considered essential for transparent reporting were not fully addressed in publications of multivariable prediction model studies.
Reporting of the items of the TRIPOD statement overall
Disclosures: C. Mendoza-Pinto, None; I. Etchegaray-Morales, None; P. Munguía-Realpozo, None; D. Osorio-Peña, None; S. Méndez-Martínez, None; M. Garcia-Carrasco, None.